摘要
影响中长期负荷的因素多,随机性强,单一预测方法很难满足不同情况的预测需要,组合预测能较好地解决单一模型的不足,但现有组合预测模型主要基于经验风险最小,预测精度受组合模型的限制。本文提出一种基于最小二乘支持向量机的中长期负荷组合预测模型,该模型利用结构风险最小化原则代替传统的经验风险最小化,充分挖掘原始数据和单一预测模型的信息,以单一模型的预测数据作为组合预测样本,选择多项式核函数的最小二乘支持向量机进行组合预测。实际算例表明,本文提出的组合模型预测平均误差仅为1.719%,具有良好的可行性和有效性。
Affected by many factors, the mid-long term load forecasting will not be satisfied if only a single method is adopted. Combination forecasting model can be used to solve the problem, but it is limited by empirical risk minimization（ERM） principle. In the paper, method based on the least square support vector machine（LS-SVM） is proposed for mid-long term load forecasting. In the model, instead of traditional ERM the principle of structure risk minimization（SRM） is used to fully mine the information of original data and the single method. The LS-SVM combination model which adopts polynomial kernel function is constructed to train the samples obtained from single methods. The simulation results show that the average error is only 1. 719% and the proposed method is feasible and effective.
出处
《电力系统及其自动化学报》
CSCD
北大核心
2008年第1期84-88,共5页
Proceedings of the CSU-EPSA
关键词
中长期负荷
组合预测
结构风险最小化
最小二乘支持向量机
预测风险
mid-long term load
combination forecasting
structure risk minimization
least square support vector machine
forecasting risk
